start adding lm_studio

src/codegate/pipeline/codegate_context_retriever/codegate.py

@@ -30,6 +30,28 @@ def name(self) -> str:
         Returns the name of this pipeline step.
         """
         return "codegate-context-retriever"
+
+    @staticmethod
+    def process_user_message(user_message: str) -> str:
+        """
+        Process a user message by extracting content inside <task> tags, if present,
+        and removing all <environment_details> tags along with their content.
+
+        Args:
+            user_message (str): The input user message.
+
+        Returns:
+            str: The processed user message.
+        """
+        # Extract content inside <task> tags if present
+        task_match = re.search(r"<task>(.*?)</task>", user_message, re.DOTALL)
+        if task_match:
+            user_message = task_match.group(1).strip()
+
+        # Remove all content inside <environment_details> tags and the tags themselves
+        user_message = re.sub(r"<environment_details>.*?</environment_details>", "", user_message, flags=re.DOTALL).strip()
+
+        return user_message    
 
     def generate_context_str(self, objects: list[object], context: PipelineContext) -> str:
         context_str = ""

src/codegate/providers/ollama/completion_handler.py

@@ -18,6 +18,7 @@ async def ollama_stream_generator(
     try:
         async for chunk in stream:
             try:
+                yield f"{chunk.model_dump_json()}\n\n"
                 # TODO We should wire in the client info so we can respond with
                 # the correct format and start to handle multiple clients
                 # in a more robust way.

src/codegate/providers/ollama/provider.py

@@ -87,7 +87,6 @@ async def create_completion(request: Request):
 
             is_fim_request = self._is_fim_request(request, data)
             try:
-                print("i create completion with data", data)
                 stream = await self.complete(data, api_key=None, is_fim_request=is_fim_request)
             except httpx.ConnectError as e:
                 logger = structlog.get_logger("codegate")
@@ -103,6 +102,4 @@ async def create_completion(request: Request):
                 else:
                     # just continue raising the exception
                     raise e
-            print("result is")
-            print(self._completion_handler.create_response(stream))
             return self._completion_handler.create_response(stream)

src/codegate/providers/openai/provider.py

@@ -1,5 +1,6 @@
 import json
 
+import httpx
 import structlog
 from fastapi import Header, HTTPException, Request
 
@@ -35,6 +36,7 @@ def _setup_routes(self):
 
         @self.router.post(f"/{self.provider_route_name}/chat/completions")
         @self.router.post(f"/{self.provider_route_name}/completions")
+        @self.router.post(f"/{self.provider_route_name}/v1/chat/completions")
         async def create_completion(
             request: Request,
             authorization: str = Header(..., description="Bearer token"),
@@ -45,7 +47,6 @@ async def create_completion(
             api_key = authorization.split(" ")[1]
             body = await request.body()
             data = json.loads(body)
-
             is_fim_request = self._is_fim_request(request, data)
             try:
                 stream = await self.complete(data, api_key, is_fim_request=is_fim_request)